Blooms of the toxic dinoflagellate Alexandrium fundyense commonly occur in the western Gulf of Maine but the amount of toxin observed in coastal shellfish is highly variable. In this study, a coupled physical-biological model is used to investigate the dynamics underlying the observed A. fundyense abundance and shellfish toxicity in 1993 (a high toxicity year) and 1994 (low toxicity year). The physical model simulates the spring circulation, while the biological model estimates the germination and population dynamics of A. fundyense based on laboratory and field data. The model captures the large-scale aspects of the initiation and development of A. fundyense blooms during both years, but small-scale patchiness and the dynamics of bloom termination remain problematic. In both cases, the germination of resting cysts accounts for the magnitude of A. fundyense populations early in the spring. Simulations with low net A. fundyense growth rates capture the mean observed concentration during the bloom peak, which is of similar magnitude during both years. There is little evidence that large-scale changes in biological dynamics between 1993 and 1994 were a primary driver of the differences in shellfish toxicity. Results instead suggest that the persistent southwesterly flow of the western Maine Coastal Current led to A. fundyense populations of similar alongshore extent by late May of both years. This period coincides with peak cell abundance in the region. Variations in wind forcing (downwelling favorable in 1993, upwelling favorable in 1994) and subsequent cell transport (inshore in 1993, offshore in 1994) in early June then provides a plausible explanation for the dramatic mid-June differences in shellfish toxicity throughout the western Gulf of Maine. ?? 2007 Elsevier Ltd. All rights reserved.